ORIGINAL ARTICLE

doi:10.1111/evo.13511

Diversification rates of the “Old Endemic” murine rodents of Luzon Island, Philippines are inconsistent with incumbency effects and ecological opportunity

Dakota M. Rowsey,1,2,3 Lawrence R. Heaney,4 and Sharon A. Jansa1,2 1Bell Museum of Natural History, University of Minnesota, St. Paul, Minnesota 55108 2Department of Ecology, Evolution, and Behavior, University of Minnesota, St. Paul, Minnesota 55108 3E-mail: [email protected] 4Field Museum of Natural History, 1400 S Lake Shore Drive, Chicago, Illinois 60605

Received April 9, 2018 Accepted May 18, 2018

Diversity-dependent cladogenesis occurs when a colonizing lineage exhibits increasing interspecific competition as it ecologically diversifies. Repeated colonization of a region by closely related taxa may cause similar effects as species within each lineage com- pete with one another. This may be particularly relevant for secondary colonists, which could experience limited diversification due to competition with earlier, incumbent colonists over evolutionary time. We tested the hypothesis that an incumbent lineage may diminish the diversification of secondary colonists in two speciose clades of Philippine “Old Endemic” murine rodents—Phloeomyini and Chrotomyini—on the relatively old oceanic island of Luzon. Although phylogenetic analyses confirm the independent, non- contemporaneous colonization of Luzon by the ancestors of these two clades, we found no support for arrested diversification in either. Rather, it appears that diversification of both clades resulted from constant-rate processes that were either uniform or favored the secondary colonists (Chrotomyini), depending on the method used. Our results suggest that ecological incumbency has not played an important role in determining lineage diversification among Luzon murines, despite sympatric occurrence by constituent species within each lineage, and a substantial head start for the primary colonists.

KEY WORDS: adaptive radiation, Phloeomyini, Chrotomyini, Murinae, oceanic island, systematics.

The factors that contribute to species and ecological diversity are genetic context, ecological opportunity may manifest as a pattern of central concern in evolutionary biology. One of the key triggers of diversity-dependent cladogenesis, in which lineage diversifica- shown to contribute to cladogenesis and morphological evolution tion rates decrease as competition for available niches increases is ecological opportunity, or the availability of unoccupied niches during a radiation (Nee et al. 1992; Rabosky and Lovette 2008; for a lineage to diversify into (Schluter 2000a; Yoder et al. 2010). Skipwith et al. 2016). The phylogenetic signature left by this pro- This opportunity can be generated by several factors, including cess can be detected by analyzing the branching patterns of the dispersal to a depauperate ecosystem, such as an oceanic island focal clade using standard diversification rate analyses (Rabosky (Baldwin and Sanderson 1998, Reddy et al. 2012; Borregaard 2006; Rabosky 2014). et al. 2017) or an evolutionary innovation that provides relief The influence of competition in shaping diversity has been from existing competition (Wainwright et al. 2012). Ecological explored for unfolding adaptive radiations in a variety of natural opportunity can facilitate adaptive radiation, rapidly producing a systems, especially islands, and previous studies have provided variety of ecologically and morphologically diverse species (Lack support for its effects on the rate of species accumulation (which 1947; Gillespie 2004; Losos 2011). When examined in a phylo- we refer to as diversification rates) and morphological change over

C 2018 The Author(s). Evolution C 2018 The Society for the Study of Evolution. 1420 Evolution 72-7: 1420–1435 DIVERSIFICATION OF LUZON OLD ENDEMIC MURINES

time (which we refer to more broadly as phenotypic evolutionary constituent species (Drake 1991). The ability of a lineage to col- rates) (Whittaker and Fernandez-Palacios 2007). For example, onize given the existence of a competitor is dependent on several Schluter and Grant (1983) showed that community assembly of factors, including habitat availability, presence of predators, and Geospiza finches is functionally overdispersed, which minimizes relatedness of invading species to incumbent species (Fukami interspecific competition for limited food resources. In a taxo- 2007; Louette and De Meester 2007; Tan et al. 2012). These nomically broad study, Schluter (2000b) used a meta-analysis short-term ecological effects on community composition, when approach to examine the prevalence of ecological character extended over large temporal scales, may result in differences in displacement, or the divergence in phenotype associated with evolutionary process between primary and secondary colonizing niche differentiation in closely related species, in natural systems. lineages. His results indicated this phenomenon is common and strongly From this macroevolutionary perspective, incumbency ef- supported in taxa such as Gasterosteus sticklebacks, Geospiza fects can be examined in two different types of systems: those finches, and heteromyid rodents of the North American desert that have been released from incumbency effects through extinc- southwest, although direct causal links between interspecific tion of competitors or dispersal to a novel environment, and those competition and phenotypic change were often lacking (Schluter in which lineages come into competition due to secondary col- 2000b). In this regard, Parent and Crespi (2009) determined onization. A popular example of diversification after extinction intraspecific variation in Galapagos´ Bulimulus land-snail shell of a competing lineage is the Cretaceous-Paleogene (K-Pg) ex- shape was negatively correlated with the number of congeners tinction and subsequent ecological release of mammals during occurring in the same habitat type and positively correlated with the Cenozoic (O’Leary et al. 2013, but see dos Reis et al. 2014). food resource heterogeneity. Finally, several recent studies of Similarly, Darwin’s finches provide a classic example of diver- lineages on oceanic islands have detected patterns of lineage sification after dispersal to a competitor-deficient system (Burns diversification consistent with declining ecological opportunity, et al. 2014). In both cases, the focal lineage undergoes rapid di- including Hawaiian leafhoppers (Bennett and O’Grady 2013), versification after existing incumbency effects—in the form of New Caledonian geckos (Skipwith et al. 2016), and Philippine competing lineages—have been removed. Detecting such exam- frogs (Blackburn et al. 2013). In sum, these studies provide ples requires analysis of diversification rates either in the context evidence that resource availability in the absence of competition of a densely sampled fossil record or with a phylogenetic per- can expand the intraspecific variation that may promote the evo- spective that includes both dispersing lineages as well as source lution of ecologically distinct species and thus result in increased taxa. Alternatively, incumbency effects on diversification may lineage diversification as well as morphological evolution. occur when two lineages come into contact through independent Nevertheless, most studies to date have focused on the evolu- colonizations of a system. In this case, we might expect sec- tion of a single lineage and have not addressed how diversification ondary colonists to experience reduced ecological opportunity— is affected when a system contains multiple, ecologically similar, and resulting lower diversification rates—compared to primary radiating lineages. Just as density dependence can slow diversi- colonists. Detecting incumbency effects in this case requires a ro- fication within a single radiating lineage, we can predict that the bust phylogenetic framework for the two interacting clades, along presence of an ecologically similar lineage may reduce the rate with analyses of diversification patterns and rates. In addition, of diversification of a second lineage that disperses to the system. researchers must also make prima facie arguments that the two If the colonizing events were asynchronous, the later-dispersing lineages are ecologically similar enough to potentially compete clade may experience a lower diversification rate proportional to over limited resources. Systems that satisfy these biological and the extent of diversification of the primary lineage at the time of methodological requirements for either approach are relatively secondary colonization. The macroevolutionary advantage held rare, meaning there is much to learn about how incumbency and by this primary colonizing lineage is an example of an incum- ecological release affect diversification. bency effect, whereby a lineage gains an advantage simply by Despite the relative scarcity of studies in which multiple being present in an area first (Case 1991; Alroy 1996). ecologically similar clades have colonized a system, several au- Whereas our study focuses on the role of incumbency at the thors have used this approach to test for incumbency effects at macroevolutionary scale, historically, incumbency (or priority) continental scales in birds, fishes, and mammals, examining both effects were hypothesized as a mechanism to explain alternative rates of lineage diversification (Betancur-R. et al. 2012; Schenk stable states in community assembly. In this context, the presence et al. 2013) and morphological evolution (Jønsson et al. 2015; of certain species can alter, whether positively or negatively, the Alhajeri et al. 2016), with mixed support for incumbency af- ability of other species to become established (Sutherland 1974). fecting diversification. However, the geographic scale at which This theory has been extended to suggest that community as- incumbency may be important in diversification remains an open sembly can be altered simply by the order of colonization of the question. For example, in a study of muroid rodents, Schenk et al.

EVOLUTION JULY 2018 1421 D. M. ROWSEY ET AL.

(2013) found some evidence for ecological opportunity affect- ing diversification at the continental scale for South American Incumbent clade, archipelago A Non-incumbent clade, muroids, but little evidence for ecological opportunity or incum- archipelago A Incumbent clade, bency effects in other geographic regions. This study did not archipelago B examine these processes at the scale of individual islands, and in some cases aggregated archipelagic and continental regions with endemic rodent assemblages (e.g., Southeast Asia), potentially masking more localized patterns of diversification. Furthermore, the analysis of continent-sized landmasses, rather than more spa-

tially limited systems, may be too broad for the detection of Net Diversification Rate adaptive radiation and/or incumbency effects at relatively short timescales. If the breadth of land area and niches available to a clade are sufficiently large or temporally dynamic, interspe- Time cific competition may be outweighed by other factors affecting diversification, such as phylogenetically conserved range limits Figure 1. Diversification rates-through-time for a three-clade sce- and climatic or geological vicariant events (Goldberg and Lande nario supporting ecological opportunity and incumbency driving diversification. A phylogenetic tree illustrates the relationships 2007; Ribas et al. 2007; Derryberry et al. 2011). As a result, the hy- among two incumbent clades and a secondary-colonizing clade. pothesis that incumbency limits ecological opportunity and slows The dotted line represents the incumbent clade in the focal sys- diversification may best be tested at finer geographic scales, with tem, the solid line represents the second colonizing clade in the clades that are more likely to be competing for niches, to mea- focal system, and the dashed line represents an incumbent clade sure its importance in determining the species diversity of natural of a different system which is sister to the secondary-colonizing systems. clade of the focal system. All clades follow a model of diversity- The endemic rodents of Luzon Island, Philippines afford the dependent cladogenesis, suggesting ecological opportunity is the opportunity to test hypotheses regarding incumbency and diver- driving diversification process, but the secondary colonist has a lower diversification rate than both the incumbent lineage and its sification rates at a spatially limited scale. The endemic murine sister lineage. rodents of Luzon have been the subject of extensive field and museum studies in recent decades (reviewed in Heaney et al. 2016a,b) and are an exceptional instance of endemism and diver- estimates of lineage diversification and phenotypic evolutionary sification following repeated colonization, with up to six colo- rates. nization events since the middle Miocene (Jansa et al. 2006). The In addition to generating the first species-level phylogeny of two oldest colonizing lineages, Phloeomyini and Chrotomyini, these two clades, we use the LOE rodents to test several predic- are referred to as the Luzon “Old Endemic” radiations and com- tions of how incumbency and ecological opportunity impact lin- prise the majority (nearly 90%) of native non-volant mammalian eage diversification using finely sampled molecular phylogenies diversity on Luzon. These two clades exhibit substantial variation and macroevolutionary inference. If ecological opportunity and in body size, diet, and other traits, both within and between each incumbency have played an important role in the diversification clade, yet species in each clade typically occur sympatrically in of LOE rodents, we first expect to recover temporally decelerating high-elevation montane and mossy forest (Heaney et al. 2016a). diversification rates in both clades consistent with clade-specific, Importantly, the Luzon Old Endemic (LOE) rodents are not sister diversity-dependent cladogenesis. Second, if the two clades share clades within Murinae and appear to have colonized Luzon sev- a background diversification rate, we expect the secondary colo- eral million years apart during the middle Miocene (Jansa et al. nizing clade (i.e., Chrotomyini) to experience a decrease in that 2006; Schenk et al. 2013; Rowe et al. 2016). An estimated four rate. Third, since the two potentially competing clades are not sis- additional rodent colonizations of Luzon have occurred within the ter clades, we expect to find a lower diversification rate in the sec- last 5 million years; these four independent colonization events ondary colonizing clade relative to its sister clade (Fig. 1). This is resulted in less diverse, primarily low-elevation clades that are col- because the secondary colonizing lineage must compete with the lectively referred to as the “New Endemic” murine rodents (Jansa primary colonizing clade as well as with other descendant species et al. 2006; Kyriazis et al. 2017). Despite extensive surveys cata- of its own lineage. Recovering support for this prediction would loging Luzon’s mammalian biodiversity, the LOE rodents lack a implicate colonization of Luzon in the decreased diversification comprehensive species-level phylogeny. The absence of this phy- rate of Chrotomyini, rather than a decreased diversification rate logeny obscures the relationships within each clade, the history inherited prior to colonization. The disparity in diversification rate of island colonization and intra-island diversification, as well as is most easily testable if the secondary-colonizing clade’s sister

1422 EVOLUTION JULY 2018 DIVERSIFICATION OF LUZON OLD ENDEMIC MURINES

lineage is a primary colonist of another system. In this case, Chro- and UMMZ04 (Smith and Patton 1991; Jansa et al. 1999); GHR: tomyini are sister to the Sahul Old Endemic (SOE) rodents, so GHRF1 (GHREXON10 in Adkins et al. 2001) and GHREND named for their range which includes New Guinea and Australia (Adkins et al. 2001); IRBP: IRBPA and IRBPB (Stanhope et al. (Rowe et al. 2008). This clade does not overlap in distribution 1996); FGB7: FGB7F1 (5’ ACGGCATGTTCTTCAGCACG 3’); with either focal clade and may exhibit ecological opportunity and FGB7R1 (5’ ATCCCTTCCAGTTCATCCACAC 3’). These consistent with an absence of potential competitors. sequence data have been submitted to the GenBank database under accession numbers MH330617-MH330660. All BRCA1, RAG1, and ACP5 sequences were obtained from previous studies. The Materials and Methods concatenated sequence matrix was deposited in TreeBase under TAXON SAMPLING accession number 22736; GenBank numbers of sequences used in We sampled 204 murid rodent species, including 39 species of our phylogenetic analyses can be found in Supporting Information Luzon Old Endemic (LOE) rodents, nine species of Old Endemic Table S1. rodents from elsewhere in the Philippine archipelago, and three All genes were amplified by PCR with a touchdown proto- species of Luzon “New Endemic” rodents to generate a com- col optimized for each locus (Jansa and Weksler 2004). Ampli- prehensive phylogeny of LOE species in the context of broader cons were sequenced using Sanger sequencing from both template Murinae. We included all species of murids currently known strands. Sequencing was performed by GENEWIZ (South Plain- from Luzon except Batomys dentatus, Crunomys fallax, and field, NJ) and the resultant reads were assembled in Geneious Tryphomys adustus. The first and second of these are known only R10 (Biomatters Ltd., Auckland, New Zealand). Consensus DNA from the holotypes, and the third from a small number of spec- sequences were aligned using MUSCLE (Edgar 2004) and con- imens without easily sampleable genetic material (Heaney et al. catenated for phylogenetic analysis. 2016a). Sequences from additional murid species were chosen pri- PHYLOGENETIC INFERENCE marily from publicly available sources obtained from previous We used PartitionFinder version 1.1.1 (Lanfear et al. 2012) to de- studies (Steppan et al. 2005; Rowe et al. 2008; Benson et al. termine the best-fitting scheme of nucleotide partitioning and cor- 2013; Schenk et al. 2013; Pages` et al. 2016; Rowe et al. 2016; responding substitution models. Our partitioning scheme varied P.-H. Fabre, Universite´ Montpellier II, personal communication, depending on the analytical approach. For maximum likelihood August 2017) and were included to place the Philippine murine analyses, we specified candidate partitions for each codon posi- radiations in the context of murid phylogeny and to provide the tion within each exon and one partition for each intron for a total best possible divergence date estimates given available fossil data. of 20 potential partitions. For the Bayesian estimation of phy- We selected taxa from subfamilies Gerbillinae, Deomyinae, and logeny, multiple partitions for each locus produced schemes that Lophiomyinae as outgroups based on their close phylogenetic prevented Markov-Chain Monte Carlo (MCMC) convergence. proximity to Murinae (Schenk et al. 2013; Pages` et al. 2016; Because of this, we used a simpler model allowing one parti- Steppan and Schenk 2017). tion for each locus, other than ACP5, for which we proposed one partition for the exons and one for the internal intron. Candidate DNA EXTRACTION AND SEQUENCING substitution models were selected based on those implemented in We sampled seven loci at varying coverage for the species RAxML (Stamatakis et al. 2008) and BEAST 2 (Bouckaert et al. in this dataset: the entire mitochondrial gene cytochrome b 2014), and compared using the Bayesian Information Criterion (CYTB, 1144 bp), exon 11 of breast cancer activating gene 1 (BIC; Schwarz 1978). (BRCA1, 2784 bp), exon 10 of growth hormone receptor (GHR, We inferred murine phylogeny using maximum likelihood in 937 bp), exon 1 of interphotoreceptor retinoid-binding protein RAxML v8.2.9, using resources of the CIPRES Science Gateway (IRBP, 1300 bp), recombination activating gene 1 (RAG1, 3040 (Stamatakis et al. 2008; Miller et al. 2010). Variants of the general bp), parts of exons 2 and 3 and the intervening intron of acid time reversible model allowing for among-site rate heterogene- phosphatase 5 (ACP5, 450 bp), and intron 7 of β-Fibrinogen ity were applied to each partition (GTR with invariant sites and (FGB7, 794 bp). gamma-distributed rates; Gu et al. 1995). We specified 10 in- For newly generated sequences, tissues were obtained from dependent searches from randomly-generated starting trees, and vouchered specimens held at the Field Museum of Natural His- nodal support was assessed using 1000 bootstrap permutations tory (FMNH) and the Smithsonian National Museum of Natural conducted using each of 10 starting trees. History (USNM). DNA was extracted using a DNeasy Blood and Bayesian estimation of phylogeny was conducted using Tissue Kit (Qiagen, Germantown, MD). We amplified genes using BEAST version 2.4.2 (Bouckaert et al. 2014) on the concate- PCR with the following primers for each locus: CYTB: MVZ05a nated data partitioned by locus using the best scheme selected

EVOLUTION JULY 2018 1423 D. M. ROWSEY ET AL.

by PartitionFinder. We linked tree models across all partitions, TESTING FOR DIVERSITY-DEPENDENT allowed site models to vary among partitions, and estimated sep- CLADOGENESIS AND DIMINISHED ECOLOGICAL arate log-normally-distributed relaxed uncorrelated clock models OPPORTUNITY for CYTB and the nuclear loci (Drummond et al. 2006). To provide To assess the role of ecological opportunity in diversification, divergence date estimates in absolute time, we specified uniform we first assessed whether the pattern of diversification was con- age priors on specific taxa as follows, based on fossil age esti- sistent with expectations of diversity-dependent cladogenesis for mates provided in Kimura et al. (2015): The “Mus-Arvicanthis” each LOE rodent clade. This model predicts initially rapid speci- split: 11.1-12.3 million years ago (Ma); the most recent com- ation that diminishes toward the present as ecological opportunity mon ancestor (MRCA) of the Arvicanthis, Otomys, and Millardia declines. We visualized lineages-through-time (LTT; Nee et al. divisions: 8.7-10.1 Ma; the MRCA of Murini: 7.3–8.3 Ma. We 1992) of the two Philippine Old Endemic clades after non-Luzon additionally applied a log-normally-distributed prior on the age species were pruned from the MCC tree to determine whether of Gerbillinae + Deomyinae with 95% quantiles of 16.0–23.0 the rate of speciation was approximately constant on a logarith- Ma (Thomas et al. 1982; Schenk et al. 2013). We specified a mic scale. Under diversity-dependent cladogenesis, lineages are Yule tree prior with an estimated birth rate given an exponen- expected to accumulate nonlinearly on a logarithmic scale, with tial prior with the initial mean set to 10. All other priors were a decrease in slope over time. This change in branching times left with default values. The BEAST analysis was conducted on was quantified using the gamma (γ) statistic applied to each LOE the CIPRES Science Gateway (Miller et al. 2010) using four clade, which tests the null hypothesis that the rate of speciation is independent runs: one run for 3.1 × 108 generations and three constant through time. Estimating a γ value less than the critical additional runs for 1.7 × 108 generations, with trees sampled value of –1.645 (at α = 0.05) indicates that speciation events are every 50,000 generations. Parameter convergence was analyzed closer to the root of the reconstructed phylogeny, and thus that using Tracer v1.6, with the first 10% of trees discarded as burn-in the majority of diversification took place earlier in the clade’s from each run before combining. The first 2.2 × 108 of the result- history, than would be expected under a pure birth diversification ing 7.4 × 108 sample generations was subsequently discarded. process (Pybus and Harvey 2000). This method assumes com- In total, 1.0 × 105 trees were generated from this remaining plete species-level sampling of the clades of interest. While we distribution. A maximum clade credibility (MCC) tree was gen- are reasonably confident that our sampling of LOE species is near erated from this posterior tree distribution using TreeAnnotator completion, our sampling of SOE rodents is far from complete. version 2.4.0 (Bouckaert et al. 2014). In addition to generating As such, we could not meaningfully perform this analysis on the concatenated trees, we generated gene trees for each locus in SOE rodents. We performed these analyses using the ape and BEAST, the MCC trees for which are reported in Supporting phytools packages in R (Paradis et al. 2004; Revell 2012; R Core Information Fig. S2. Team 2015). We also tested for temporally decelerating diversification rates using a model-fitting approach. We compared the fit of five ESTIMATES OF LUZON COLONIZATION TIMES evolutionary models, including a Yule diversification model, a We used the Bayesian posterior tree distribution to estimate col- constant-rate birth-death model, and time-dependent rate models onization times for each of the five Luzon rodent groups we with decreasing speciation rates (SPVAR), increasing extinction sampled. For the LOE rodents, we calculated the interval of colo- rates (EXVAR), or both (BOTHVAR; Yule 1924; Nee et al. 1994; nization as the time between the stem and crown ages of the clade, Rabosky 2006). The fit of these models was compared using corresponding to the maximum and minimum colonization times the Akaike information criterion corrected for small sample sizes for each clade. These times were calculated across the 95% high- (AICc; Akaike 1974; Hurvich and Tsai 1989), with the model with est posterior density (HPD) of divergence times for each clade. the highest Akaike weight (w) accepted as the best. Model fitting We generated probability distributions by populating 0.2 My bins was conducted using the laser package of R using default pa- between the maximum and minimum age for each estimate with rameter settings for each function (Rabosky 2006). To ensure that the normalized proportion of trees in which the given date fell model selection was not driven by the choice of a Yule tree prior in between the stem and crown ages of a clade. Because intra-island our BEAST analysis, we performed an additional model selection speciation events were unavailable to sample for the three New analysis using a distribution of trees sampled under a birth–death Endemic rodent colonization events, we calculated the maximum prior with the topology constrained to that of the original tree colonization ages for these species as the divergence date be- (samples generated: 2.9×108, trees generated: 2.9×105). As with tween each Luzon species and their non-Luzon sister lineage. the LTT analysis, we did not perform diversification model fit- The distribution of times for each clade was plotted to visualize ting using on the SOE rodents, for which we lack species-level the probability of independent colonization. sampling.

1424 EVOLUTION JULY 2018 DIVERSIFICATION OF LUZON OLD ENDEMIC MURINES

To test for incumbency effects, we assessed whether the sec- Table 1. Partitions best supported by model comparison using ondary colonizing clade (Chrotomyini) has a lower diversifica- PartitionFinder version 1.1.1 (Lanfear et al. 2012). tion rate than either (1) the incumbent clade that occupies the Partition Model Coding positions same system (Phloeomyini) as well as (2) its sister clade (the SOE), which diversified across Sahul in the absence of an earlier- RAxML 1 GTR+I+ CYTB pos 1 arriving clade of murine competitors. We estimated diversifica- RAxML 2 GTR+I+ CYTB pos 2 tion rate parameters for these three, independent radiations using RAxML 3 GTR+I+ CYTB pos 3 + + a likelihood-based framework that accounts for lineage-specific RAxML 4 GTR I GHR pos 1, IRBP pos 2, RAG1 pos 1, ACP5 sampling probabilities (Alfaro et al. 2009; estimated using modi- exon pos 2 fied MEDUSA code from Jansa et al. 2014). For this analysis, we RAxML 5 GTR+I+ GHR pos 2, IRBP pos 3, pruned our MCC tree to include the two focal groups in each test. RAG1 pos 2, ACP5 Furthermore, we pruned the SOE clade to genus level (i.e., by re- exon pos 3 moving Pogonomys macrourus and P. sylvestris) and enumerated RAxML 6 GTR+ GHR pos 3, BRCA1 pos the species represented by each genus according to Musser and 3, FGB7 Carleton (2005) supplemented by Rowe et al. (2008) and Rowe RAxML 7 GTR+ IRBP pos 1, RAG1 pos 3 + et al. (2016). Pruning two of the three species of Pogonomys was RAxML 8 GTR BRCA1 pos 1, BRCA1 pos 2, ACP5 exon pos 1 necessary due to incomplete phylogenetic sampling of this genus RAxML 9 GTR+I+ ACP5 intron and uncertainty associated with relationships of these unsampled BEAST 1 GTR+I+ CYTB species to the three species that were sampled. We then estimated BEAST 2 K80+I+ GHR the maximum likelihood parameter values and 95% confidence BEAST 3 SYM+I+ IRBP interval on the net diversification rate (r = λ – μ,whereλ repre- BEAST 4 GTR+ BRCA1 sents speciation rate and μ represents extinction rate) and extinc- BEAST 5 SYM+I+ RAG1 tion fraction (ε = μ/λ) for each clade. This confidence interval BEAST 6 SYM+I+ ACP5 exon + + was approximated by generating the likelihood surface up to 3 BEAST 7 HKY I ACP5 intron BEAST 8 GTR+ FGB7 log-likelihood units from the maximum likelihood estimate of these parameters. If incumbency has affected diversification of Chrotomyini, we would expect this clade to exhibit a lower net We used initial parameters suggested by the setBAMMPriors diversification rate than both Phloeomyini (the competing clade) function of the R package BAMMtools (Rabosky et al. 2014) and and the SOE rodents (its sister clade, which radiated elsewhere conducted an MCMC analysis for 108 generations, saving one in without other murine competitors). Statistical significance for the every 1000 states. We discarded 10% of samples of the resultant difference in rates is inferred if the maximum likelihood estimate distribution as burn-in and subsampled the remaining samples to (MLE) of diversification rate for one clade falls outside an inter- 5000. We then examined the 95% credible shift set of this dis- val described by 3 log-likelihood units around the MLE for the tribution to determine support for rate shifts on branches leading other. to the clades containing LOE rodents. We interpreted support Finally, as an additional approach to testing these three di- for diversification driven by ecological opportunity as recovering versification rate hypotheses, we examined rate heterogeneity diversity-dependent cladogenesis with a higher initial diversifica- using Bayesian Analysis of Macroevolutionary Mixtures v2.5.0 tion rate in the LOE rodent clades compared to the background (BAMM; Rabosky 2014) to test differences in diversification rate diversification rate. Conversely, we interpreted support for dimin- for each clade of LOE rodents, while allowing for time-dependent ished ecological opportunity as recovering a rate shift followed diversification rates and clade-specific sampling probabilities. We by decreasing diversification rates along the branch leading to used a backbone species sampling probability of 0.89. This value Chrotomyini. was calculated as the ratio of the estimated number of species within murine genera sampled in our phylogenetic analysis out of the total number of murine species, effectively accounting for Results the proportion of murine rodents in genera unsampled in our phy- PHYLOGENETIC RELATIONSHIPS AMONG MURINES logenetic analysis. We also determined genus-specific sampling The PartitionFinder analysis identified nine nucleotide partitions probabilities to account for unsampled species within sampled for maximum likelihood (ML) analysis and eight for the Bayesian genera. In both cases, murid species richness was determined analysis as the best fit models given the candidate partitions using Musser and Carleton (2005), supplemented by additional (Table 1). Among the OE rodents, bootstrap support values in ML sources for many genera (Supporting Information Table S2). analysis were comparatively weakest within the genus Apomys,

EVOLUTION JULY 2018 1425 D. M. ROWSEY ET AL.

A Acomys ignitus B Chiropodomys gliroides Uranomys ruddi Apomys camiguinensis Tatera robusta Apomys hylocoetes Gerbillus gerbillus Apomys insignis Meriones shawi Apomys musculus Lophiomys imhausi Apomys microdon Hapalomys delacouri R 5237 Apomys gracilirostris Hapalomys delacouri ZIN 98922 Apomys minganensis Phloeomys pallidus Phloeomyini Apomys abrae Phloeomys cumingi Apomys datae Crateromys australis Apomys banahao Batomys hamiguitan Apomys brownorum Chrotomyini Batomys salomonseni Apomys sacobianus Crateromys heaneyi Apomys lubangensis Crateromys schadenbergi To A Apomys sierrae Batomys granti Apomys magnus Batomys uragon Apomys aurorae Carpomys melanurus Apomys iridensis Carpomys phaeurus Apomys zambalensis Musseromys gulantang Archboldomys luzonensis Musseromys inopinatus Archboldomys maximus Musseromys anacuao Rhynchomys banahao Musseromys beneficus Rhynchomys isarogensis Micromys minutus Rhynchomys soricoides Maxomys bartelsii Rhynchomys tapulao Maxomys moi Soricomys musseri Maxomys surifer Soricomys leonardocoi Maxomys cf hellwaldii Soricomys kalinga Maxomys dollmani Soricomys montanus Maxomys hellwaldii Chrotomys silaceus Crunomys celebensis Chrotomys sibuyanensis Crunomys melanius Chrotomys gonzalesi Crunomys suncoides Chrotomys mindorensis Maxomys musschenbroekii Chrotomys whiteheadi Maxomys hylomyoides Anisomys imitator Maxomys whiteheadi Chiruromys vates Maxomys panglima Lorentzimys nouhuysi Maxomys pagensis Macruromys major Maxomys rajah Hyomys goliath Sommeromys macrorhinos Pogonomys loriae Hyorhinomys stuempkei Pogonomys macrourus Waiomys mamasae Pogonomys sylvestris Echiothrix centrosa Mammelomys lanosus Paucidentomys vermidax Abeomelomys sevia Melasmothrix naso Mallomys rothschildi Tateomys macrocercus Leggadina forresti Tateomys rhinogradoides Leporillus conditor Saxatilomys paulinae Conilurus penicillatus SOE Dacnomys millardi Mesembriomys gouldii Leopoldamys sabanus Zyzomys argurus Leopoldamys edwardsi Pseudomys australis Leopoldamys neilli Notomys fuscus Margaretamys elegans Mastacomys fuscus Chiromyscus chiropus Uromys caudimaculatus Niviventer cremoriventer Paramelomys levipes Niviventer excelsior Solomys salebrosus Niviventer confucianus Melomys rufescens Niviventer culturatus Crossomys moncktoni Srilankamys ohiensis Parahydromys asper Berylmys berdmorei Hydromys chrysogaster Berylmys bowersi Leptomys elegans Bullimus gamay Paraleptomys sp Bullimus bagobus Xeromys myoides Bullimus luzonicus Pseudohydromys ellermani Sundamys muelleri Microhydromys richardsoni Bunomys chrysocomus Millardia kathleenae Paruromys dominator Otomys anchietae Abditomys latidens Otomys denti Diplothrix legata Otomys angoniensis Bandicota bengalensis Parotomys brantsii Bandicota savilei Golunda elliotti Rattus exulans Micaelamys namaquensis Rattus leucopus Grammomys macmillani Rattus verecundus Grammomys dolichurus Rattus giluwensis Grammomys ibeanus Rattus novaguineae Dasymys incomtus Rattus praetor Mylomys dybowskii Rattus sordidus RTL Rhabdomys pumilio Rattus villosissimus Arvicanthis neumanni Rattus everetti Arvicanthis niloticus Tarsomys apoensis Lemniscomys barbarus Limnomys bryophilus Lemniscomys striatus Limnomys sibuanus Oenomys hypoxanthus Rattus norvegicus Hybomys univittatus Rattus tiomanicus Dephomys defua Rattus rattus Stochomys longicaudatus Rattus tanezumi Vandeleuria oleracea Malacomys longipes

To B Mus pahari Bootstrap support: Mus cervicolor Mus cookii : 100% Mus caroli Mus booduga : 90-99% Mus terricolor Mus musculus : 75-89% Mus spicilegus Tokudaia osimensis Apodemus mystacinus All other nodes blank Apodemus sylvaticus Apodemus flavicollis Apodemus speciosus 0.04 sub/site Apodemus agrarius Apodemus latronum Apodemus semotus Praomys degraaffi Praomys jacksoni Praomys misonnei Praomys tullbergi Myomyscus brockmani Stenocephalomys albipes Mastomys erythroleucus Mastomys hildebrandtii Colomys goslingi Zelotomys hildegardeae Heimyscus fumosus Hylomyscus parvus Hylomyscus stella

Figure 2. Maximum likelihood tree topology generated from concatenated sequence data. Branches are proportional to number of nucleotide substitutions. Dots at nodes correspond to bootstrap support (BS): BS = 100%: black, 90% ࣘ BS < 100%: gray, 75% ࣘ BS < 90%: white, BS < 75%: blank. Clades containing Luzon-endemic murines are enclosed in gray boxes. Sahul Old Endemics (SOE) and Rattus- Tarsomys-Limnomys (RTL) are abbreviated for clarity. with other areas of weak support including divergences within Crateromys australis, a species endemic to Dinagat Island and Chrotomys, Musseromys,andSoricomys (Fig. 2). Bayesian nodal represented only by the species holotype, was placed in a well- support values were generally stronger than their ML counterparts supported clade containing the other two species of Greater Min- (>0.95) apart from some areas of weak support within Chrotomys, danao phloeomyines, Batomys salomonseni and Batomys hami- Musseromys,andSoricomys (Fig. 3). guitan. The remaining Batomys and Crateromys species found on The phylogenetic relationships we recovered among OE ro- Luzon and Panay have a robust sister relationship (Figs. 2 and 3). dents were generally consistent with previous studies (Jansa et al. Among relevant non-Philippine Asian murines, the phylo- 2006; Schenk et al. 2013; Justiniano et al. 2015; Rowe et al. 2016). genetic placement of Vandeleuria and Millardia remains un- A single exception occurred in the placement of Apomys minga- certain. Our study corroborates previous molecular studies that nensis, which we recovered as sister to Apomys abrae and Apomys fail to find the suggested alliance of Vandeleuria oleracea with datae compared to the placement as sister to the clade containing Chiropodomys that has been postulated based on dental charac- A. brownorum and A. sacobianus as estimated in previous work ters (Musser and Carleton 2005) and instead strongly supports (Justiniano et al. 2015). Interestingly, the phloeomyine genera a sister-taxon relationship between Chiropodomys and the clade Crateromys and Batomys were not reciprocally monophyletic. containing Chrotomyini + SOE murines (Schenk et al. 2013;

1426 EVOLUTION JULY 2018 DIVERSIFICATION OF LUZON OLD ENDEMIC MURINES

A Lophiomys imhausi B Millardia kathleenae Acomys ignitus Chiropodomys gliroides Uranomys ruddi Apomys microdon L/O Gerbilliscus robustus Apomys musculus L/O A Gerbillus gerbillus Apomys camiguinensis M Meriones shawi Apomys hylocoetes M Hapalomys delacouri R5237 Apomys insignis M

Hapalomys delacouri ZIN98922 To A Apomys gracilirostris O

Phloeomys cumingi L Phloeomyini Apomys minganensis L Phloeomys pallidus L Apomys abrae L Batomys granti L Apomys datae L L L Batomys uragon Apomys banahao Chrotomyini Crateromys heaneyi P Apomys brownorum L Crateromys schadenbergi L Apomys lubangensis U Crateromys australis D Apomys sacobianus L Batomys hamiguitan M Apomys sierrae L Batomys salomonseni M Apomys magnus L Carpomys melanurus L Apomys aurorae L Carpomys phaeurus L Apomys iridensis L Musseromys inopinatus L Apomys zambalensis L Musseromys gulantang L Archboldomys luzonensis L Musseromys anacuao L Archboldomys maximus L Musseromys beneficus L Rhynchomys banahao L Micromys minutus Rhynchomys isarogensis L Maxomys bartelsii Rhynchomys soricoides L Maxomys moi Rhynchomys tapulao L Maxomys surifer Chrotomys sibuyanensis S Maxomys panglima Chrotomys silaceus L Maxomys pagensis Chrotomys whiteheadi L Maxomys rajah Chrotomys gonzalesi L Maxomys cf hellwaldii Chrotomys mindorensis L/O Maxomys musschenbroekii Soricomys leonardocoi L Maxomys hylomyoides Soricomys musseri L Maxomys whiteheadi Soricomys kalinga L Crunomys celebensis Soricomys montanus L Crunomys melanius Anisomys imitator Crunomys suncoides Chiruromys vates Maxomys dollmani Lorentzimys nouhuysi Maxomys hellwaldii Macruromys major Sommeromys macrorhinos Hyomys goliath Hyorhinomys stuempkei Pogonomys loriae Waiomys mamasae Pogonomys macrourus Echiothrix centrosa Pogonomys sylvestris Paucidentomys vermidax Mammelomys lanosus Tateomys rhinogradoides Abeomelomys sevia Melasmothrix naso Mallomys rothschildi Tateomys macrocercus Leggadina forresti Saxatilomys paulinae Leporillus conditor Margaretamys elegans Conilurus penicillatus SOE Chiromyscus chiropus Mesembriomys gouldii Niviventer cremoriventer Zyzomys argurus Niviventer excelsior Pseudomys australis Niviventer confucianus Mastacomys fuscus Niviventer culturatus Notomys fuscus Dacnomys millardi Uromys caudimaculatus Leopoldamys sabanus Paramelomys levipes Leopoldamys edwardsi Melomys rufescens Leopoldamys neilli Solomys salebrosus Srilankamys ohiensis Crossomys moncktoni Berylmys berdmorei Hydromys chrysogaster Berylmys bowersi Parahydromys asper Bullimus gamay Leptomys elegans Bullimus bagobus Paraleptomys sp Bullimus luzonicus L Xeromys myoides Sundamys muelleri Microhydromys richardsoni Bunomys chrysocomus Pseudohydromys ellermani Paruromys dominator Vandeleuria oleracea Abditomys latidens L Otomys angoniensis Diplothrix legata Parotomys brantsii Bandicota bengalensis Otomys anchietae Bandicota savilei Otomys denti Rattus everetti L Golunda elliotti Tarsomys apoensis C Micaelamys namaquensis Limnomys bryophilus Grammomys macmillani Limnomys sibuanus Grammomys dolichurus Rattus giluwensis Grammomys ibeanus

Rattus novaguineae RTL Oenomys hypoxanthus Rattus praetor Hybomys univittatus Rattus leucopus Dephomys defua Rattus verecundus Stochomys longicaudatus Rattus sordidus Dasymys incomtus Rattus villosissimus Mylomys dybowskii Rattus exulans Rhabdomys pumilio Rattus norvegicus Lemniscomys barbarus Rattus tiomanicus Lemniscomys striatus

To B Rattus rattus Arvicanthis neumanni Rattus tanezumi Arvicanthis niloticus B Tokudaia osimensis Apodemus mystacinus Apodemus flavicollis Apodemus sylvaticus Apodemus speciosus Apodemus agrarius Apodemus latronum Apodemus semotus Malacomys longipes Mus pahari Mus booduga Mus terricolor Mus musculus D Mus spicilegus Mus caroli Mus cervicolor Mus cookii Colomys goslingi Zelotomys hildegardeae Myomyscus brockmani Stenocephalomys albipes Mastomys erythroleucus Mastomys hildebrandtii Praomys degraaffi Praomys jacksoni Praomys misonnei Praomys tullbergi Heimyscus fumosus Hylomyscus parvus Hylomyscus stella

Figure 3. Time-calibrated maximum clade credibility (MCC) tree generated from concatenated sequence data. Nodes with ࣙ0.95 PP are indicated with a dot. Bars at nodes represent 95% highest posterior density (HPD) intervals for node ages. Nodes constrained with fossil data (Thomas et al. 1982; Kimura et al. 2015) are indicated with letters circumscribed by black circles. Clades containing Luzon- endemic murines are enclosed in gray boxes. Letters adjacent to taxon names correspond to the endemic faunal region of the species. L: Luzon Island, P: Panay I., D: Dinagat I., M: Mindanao I., O: Mindoro I., S: Sibuyan I., U: Lubang I. Sahul Old Endemics (SOE) and Rattus-Tarsomys-Limnomys (RTL) are abbreviated for clarity.

Steppan and Schenk 2017). Otherwise, the phylogenetic position of 3.6 ± 0.4, 2.8 ± 0.3, and 1.4 ± 0.2 Ma for the lineages Abdito- of Vandeleuria is not secure in any molecular analysis to date mys latidens, Rattus everetti,andBullimus luzonicus, respectively. (Schenk et al. 2013; Pages` et al. 2016; this study). Likewise, No overlap in distribution of colonization times occurred between the placement of Millardia (represented here by M. kathleenae) the Old Endemic and New Endemic groups (Fig. 4). The coloniza- varies depending on analytical approach (Fig. 3; see also Schenk tion ages between Phloeomyini and Chrotomyini exhibited only et al. 2013), and is also not well supported by any analysis of any slight overlap: 3.6% of the phloeomyine 95% HPD overlappped molecular dataset to date. with 7.4% of the chrotomyine density, for a pooled probability density of 5.2%, suggesting time-staggered colonization between the two Old Endemic clades was likely (Fig. 4). ESTIMATES OF LUZON COLONIZATION TIMES Median estimates of colonization time for the LOE rodents were 12.8 ± 1.2 and 8.4 ± 0.9 Ma for Phloeomyini and Chrotomyini, DIVERSITY-DEPENDENT CLADOGENESIS AND respectively, with a median difference in crown clade age of DIMINISHED ECOLOGICAL OPPORTUNITY 3.9 ± 0.9 My, suggesting Phloeomyini were present on Luzon LTT plots and γ−statistics for the Luzon representatives of at least 3 My prior to Chrotomyini. We obtained mean maximum each Old Endemic clade (i.e., excluding taxa occurring on other colonization age estimates of Luzon for the New Endemic rodents Philippine islands) failed to show any signal of temporally

EVOLUTION JULY 2018 1427 D. M. ROWSEY ET AL.

0.35 Phloeomyini did not differ when conducted using an MCC tree generated un- Chrotomyini Abditomys latidens = = Rattus everetti der a birth–death tree prior (ln(L)P –10.33, AICcP 23.10, 0.30 Bullimus luzonicus wP = 0.78; ln(L)C = 12.83, AICcC = –23.21, wC = 0.78).

0.25 Likelihood estimation of diversification rate parameters for the two LOE clades corroborated the differences in rate parame- 0.20 ters obtained using our model selection approach, with a differ- ence in maximum likelihood estimates (MLE) >3 log-likelihood 0.15

Relative density units for each clade (Fig. 6A, Table 2). Rate parameter compar- 0.10 isons for Chrotomyini and SOE rodents, by contrast, suggested the two clades evolved under processes with comparable diver- 0.05 sification rates and similarly low extinction fractions (Fig. 6B,

0 Table 2). The MLE for all clades under a birth–death process was 18 16 14 12 10 8 6 4 2 0 Bayesian 95% posterior age estimate (My) not significantly different from the maximum likelihood assuming aYuleprocess(DP = 0.88, P = 0.35, df = 1; DC = 0.036, p = 0.85, Figure 4. Histogram of inferred colonization ages for each clade df = 1; DS = 0.050, P = 0.82, df = 1). Both comparisons failed of murine colonists on Luzon Island. Bars represent cladewise to support the hypotheses that diversification of Chrotomyini probability of colonization during 0.2 My intervals. Intervals for was depressed with respect to either Phloeomyini or the SOE Phloeomyini and Chrotomyini represent 95% HPD of minimum and maximum age estimates inferred from crown and stem clade ages rodents. respectively. Intervals for Abditomys latidens, Rattus everetti,and Analysis of diversification rates using BAMM suggests Bullimus luzonicus are inferred from ages of divergence from their nearly constant diversification rates for the majority of Muri- non-Luzon sister taxa and thus represent maximum colonization nae (Fig. 7). We recovered strong support for a model allowing ages. rate heterogeneity in the murine tree, with the largest posterior- to-prior ratio supporting three rate shifts (Bayes factor: 108.8, but such models were rarely sampled). Instead, both the maximum a posteriori (MAP) rate-shift configuration and modal number of

γ 5.42×10-2 (ns) P: Phloeomyini shifts inferred in our BAMM analysis suggest two diversification γ -1 C: 1.01×10 (ns) Chrotomyini rate shifts. These rate shifts were proposed in similar locations

10 20 across the 95% credible-shift set. First, we recovered an increas- ing extinction rate along the branch leading to Lophiomys imhausi 5 (the sole representative of Lophiomyinae), yielding a lower net diversification rate compared to the rest of Muridae. Second, an Number of lineages

2 additional shift with increased speciation and extinction rates oc- curredalongthebranchleadingtotheRattus-Tarsomys-Limnomys

1 (RTL) division, yielding an increased net diversification rate (Sup- 12 10 8 6 4 2 0 My before present porting Information Fig. S2). The remainder of the tree exhibits a slowly declining net diversification rate with near-zero extinc- Figure 5. Lineage-through-time plots for Luzon Phloeomyini and tion, with an average time-integrated diversification rate of 0.30 Chrotomyini. The branching times for each clade could not be dis- lineages/My across the posterior distribution. In contrast to the tinguished from that of a constant-rate process. nearly twofold difference in rates inferred from the ML analyses reported above, this analysis recovers nearly identical average di- -2 dynamic diversification rates (Phloeomyini: γP = 5.42×10 , versification rates for Phloeomyini and Chrotomyini (rP = 0.29, -1 P = 0.96; Chrotomyini: γC = 1.01×10 , P = 0.91; Fig. 5). rC = 0.28; Table 2). The 95% credible-shift set included 38 distinct Furthermore, we recovered a Yule diversification process, rather rate shift configurations, but were largely similar to one another than time-variable processes, as the best-fit model of lineage and the MAP configuration. Rate shifts were frequently sampled diversification for each LOE group (Phloeomyini: ln(L)P = along the split between Lophiomys and remaining Muridae, with

–10.27, AICcP = 22.99, wP = 0.78; Chrotomyini: ln(L)C = 15.20, a lower net diversification rate than the remaining Muridae, and

AICcC = –27.96, wC = 0.77). Also contrary to our predictions, in the core Rattus division, with higher diversification rates than we recovered a net diversification rate (r) of Luzon Chrotomyini the background diversification process (Supporting Information more than double that of Luzon Phloeomyini using this model Fig. S3). Consistent with our maximum likelihood analyses of selection approach (rP = 0.16, rC = 0.41). The best-fit model diversification rates (Fig. 6), we failed to recover support for

1428 EVOLUTION JULY 2018 DIVERSIFICATION OF LUZON OLD ENDEMIC MURINES

Figure 6. Comparison of the net diversification rate (r = λ – μ) and extinction fraction (ε = μ/λ) parameters inferred using MEDUSA for each of three murine clades. Panel (A) compares the Luzon Chrotomyini (orange, solid lines, filled point) and Phloeomyini (blue, dashed lines, empty circle), whereas 6B compares the Luzon Chrotomyini with the Sahul Old Endemic (SOE) rodents (green, dashed lines, empty circle). Higher likelihood values are represented with brighter colors. The variation in the size of the likelihood surface stems from variation in uncertainty of parameter estimates for each clade. The maximum likelihood estimate (MLE) for each clade is shown with a point and –1, –2, and –3 log-likelihood unit contours are shown with lines. The parameter area within three log-likelihood units approximates a 95% confidence interval of each clade’s MLE, and recovering a MLE outside of the interval of the other clade suggests that the two exhibit distinct evolutionary processes.

Table 2. Diversification rate parameters inferred for Phloeomyini, Chrotomyini, and Sahul Old Endemic (SOE) rodents using MEDUSA and BAMM.

MEDUSA BAMM

Lineages/My Phloeomyini Chrotomyini SOE Phloeomyini Chrotomyini SOE

Speciation rate (λ)3.2× 10–1 4.2 × 10–1 4.2 × 10–1 3.0 × 10–1 2.9 × 10–1 3.0 × 10–1 Extinction rate (μ) 2.3 × 10–1 6.3 × 10–2 4.6 × 10–2 1.0 × 10–2 1.0 × 10–2 8.8 × 10–3 Diversification rate (r)8.1× 10–2 3.6 × 10–1 3.7 × 10–1 2.9 × 10–1 2.8 × 10–1 2.9 × 10–1 Extinction fraction (ε) 0.74 0.14 0.11 3.4 × 10–2 3.5 × 10–2 3.0 × 10–2 our hypotheses of lowered diversification rates in Chrotomyini Discussion compared to either Phloeomyini or the SOE rodents. Although ESTIMATES OF LUZON COLONIZATION TIMES we recovered slowly declining diversification rates for both LOE In reconstructing the first species-level phylogeny of LOE ro- clades, the generating process is identical for most of the remain- dents, we provide evidence that colonization of Luzon by these der of the murine tree. two major rodent groups was temporally staggered. Previously,

EVOLUTION JULY 2018 1429 D. M. ROWSEY ET AL.

Background fit for the diversification of the two clades of LOE rodents, but Phloeomyini Chrotomyini failed to recover any evidence of temporally decelerating cladoge- A B nesis. Our BAMM analysis recovered slightly decreasing diversi- fication rates through time for both Chrotomyini and Phloeomyini, but this process appears to be no different from the remainder of Murinae (excluding Rattus-Tarsomys-Limnomys,whichshow accelerated rates of diversification). Based on these results, we conclude that species diversification of LOE rodents is not con- sistent with expectations of ecological opportunity presented by colonizing a new landmass.

Net diversification rate The second prediction we tested was that Chrotomyini should exhibit lower diversification rates than Phloeomyini, due to in- cumbency effects. Two lines of evidence suggest that this was not the case. First, we did not detect a change in diversification rate

My before present in the murine phylogeny between these two clades in our BAMM analysis (Fig. 7). Second, although we did recover a significant Figure 7. Diversification rate-through-time of Murinae and out- difference in diversification rate using maximum-likelihood anal- groups inferred from the distribution of BAMM rate configura- ysis, this difference was not in the expected direction: the esti- tions. Dashed line represents net diversification rate for the entire mated diversification rate of Chrotomyini in these analyses was tree inferred from the average of the posterior (time-integrated rate: 0.30 lineages/My); gray envelope indicates the 95% confi- up to fourfold faster than that of Phloeomyini (Fig. 6, Table 2). dence interval of the estimate. The dotted gray line indicates Recovering, at minimum, similar diversification rates, or as we average diversification rate inferred for Phloeomyini (0.29 lin- found, a much higher diversification rate in Chrotomyini, directly eages/My) and the solid gray line indicates the inferred rate of contradicts the expectation that a colonizing clade is prevented Chrotomyini (0.28 lineages/My), with circles at the clade origin for from achieving comparable diversity to an incumbent lineage. clarity. Lettered circles indicate inferred rate shift positions based Our final prediction was that Chrotomyini should exhibit a on the maximum a posteriori (MAP) rate shift configuration: (A) ex- lower diversification rate than their sister clade, the Sahul Old En- tinction rate increase along branch leading to Lophiomys imhausi; demics, which constitutes the earliest rodent colonists of the Sahul (B) speciation and extinction rate increase along branch leading to the Rattus-Tarsomys-Limnomys (RTL) division. region. Comparing these two clades allowed us to explore how chrotomyine diversification may have unfolded in the absence of primary colonists on Luzon. Again, in contrast to our predictions, a two-phase model for colonization of Philippine Old Endemic the two clades had diversification rates consistent with evolution versus New Endemic rodents was supported, but individual clades under a single macroevolutionary process. The uniformity we re- within each of these groups could not be temporally differentiated covered provides an additional line of evidence that Chrotomyini (Jansa et al. 2006). Our study refines this estimate by temporally were not limited in their diversification by colonizing Luzon after separating the invasion of Luzon by Phloeomyini from that of Phloeomyini. Our failure to recover support for any of the three Chrotomyini, and provides the first step in differentiating the tim- predictions that formed the basis of this study point to the con- ing of colonization of the Luzon New Endemics, suggesting inde- clusion that the diversification history of the Luzon Old Endemic pendent colonization of Bullimus luzonicus from Abditomys lati- rodents is more prominently influenced by factors other than in- dens and Rattus everetti (based on maximum clade ages; Fig. 4). cumbency effects. The dominant pattern for the LOE rodents, Recovering Phloeomyini and Chrotomyini as colonizing millions and the majority of Murinae, is a constant or slowly decelerating of years apart provides the basis for the prediction that compe- diversification rate with a relatively low apparent extinction rate. tition for niches was potentially weaker for the phloeomyines The patterns illustrated in these two clades of island rodents compared to the chrotomyines. corroborate the patterns found for rodents in continental systems (Schenk et al. 2013). These authors found that incumbency ef- DIVERSITY-DEPENDENT CLADOGENESIS AND fects, whereby secondary colonists exhibit diminished diversifi- DIMINISHED ECOLOGICAL OPPORTUNITY cation rates compared to primary colonists, were the exception For this study, we sought to test three predictions regarding how rather than the rule among muroid rodents. The authors recov- ecological opportunity, diminished in secondary colonists by in- ered diversity-dependent cladogenesis among South American cumbency of another clade, alters lineage diversification. First, we muroid rodents with strong support, and among the SOE rodents tested whether diversity-dependent cladogenesis provided a good with weak support, but found no support for secondary colonists

1430 EVOLUTION JULY 2018 DIVERSIFICATION OF LUZON OLD ENDEMIC MURINES

having decreased diversification rates in any of their biogeo- 2016a). This temporal dynamism may mean speciation is driven graphic categories. In aggregate, these findings suggest that in- less by competition for niches and more by dispersal to newly cumbency’s influence on lineage diversification may be eas- available habitat generated by geological processes, specifically, ily overshadowed by other phenomena (Derryberry et al. 2011; the montane “islands” formed by volcanism and tectonic uplift Schenk et al. 2013). With this context in mind, there are sev- over the past 15 million years. Importantly, these montane is- eral mutually compatible scenarios that could be responsible for lands, while originally isolated from one another in the form of the observed diversification history of the LOE. We outline three individual oceanic islets, probably coalesced into their contem- below. porary configuration only within the last 5 million years (Heaney First, Phloeomyini may have not had enough time for suffi- et al. 2016a). Considering many of the species-level divergences cient speciation to occur to impose incumbency effects on Chro- in both LOE clades began after Luzon’s coalescence, and that sis- tomyini. To assess how much diversification could have occurred ter species within either clade often occupy nonadjacent ranges within Phloeomyini before Chrotomyini arrived on Luzon, we (Heaney et al. 2016b), we suggest that a mixture of over-land and simulated the number of phloeomyine lineages present at the in- over-water dispersal to other mountain ranges on Luzon has been ferred median colonization age of Chrotomyini. Based on the di- at least as important in LOE evolutionary history as processes versification rates inferred using MEDUSA and BAMM, we per- of ecological specialization and differentiation, and that lowland formed 100 constant-rate birth–death simulations, which yielded habitat presents a prominent, but not insurmountable, dispersal an average of three species present for each set of assumed rates barrier to most LOE species. The patterns of allopatric speciation (simulations performed using TESS in R; Hoehna 2013). The rela- we propose are supported in several clades of Southeast Asian tively low rate of early cladogenesis among Phloeomyini is further mammals (Esselstyn et al. 2009; Achmadi et al. 2013; Justiniano supported by our failure to recover diversity-dependent cladogen- et al. 2015), suggesting that the climatic and geological dynamism esis as the best-supported diversification model for this clade. Ad- of this region may be paramount in explaining its biodiversity. ditionally, if density-independent time-for-speciation processes Third, the effects of incumbency may still be apparent in were the dominant mechanism of early diversification in this this system, but they are acting not on diversification history, but clade, as suggested by our analyses, then we would not expect rather on patterns of phenotypic diversity. If true, we may still the rapid early cladogenesis predicted under a model of ecolog- recover diversity-dependent trait evolution with decreased rates ical opportunity (Stephens and Wiens 2003; Rabosky 2012). As of phenotypic evolution among secondary colonists. This hypoth- a result, the relatively low early diversity of phloeomyine species esis seems intuitive, considering phenotypic evolution driven by may not have exhibited a strong enough competitive pressure on ecological opportunity forms the basis for adaptive radiation the- the secondary colonists to depress their diversification. ory (Parent and Crespi 2009; Mahler et al. 2010), but it remains Second, the extent of geographic dynamism may be too great to be tested in this system. In the context of incumbency, the to impose realistic ecological limits on species diversity, even at rate of morphological evolution may be suppressed in secondary the scale of a single island. Numerous studies encompassing a colonists, although lineage diversification (via allopatric specia- variety of clades illustrate that allopatric speciation is a prevalent tion) is unaffected, yielding many ecologically similar species. mode of diversification on continental landmasses (Ribas et al. Alternatively, incumbent clades may filter secondary colonists 2007; Derryberry et al. 2011; Giarla and Jansa 2014). At large, ge- in favor of those lineages occupying distinct regions of ecomor- ographically complex spatial scales, vicariance or dispersal events phospace. The resulting morphological variation may be such that facilitate diversification without increasing interspecific competi- cladewise overlap is minimized (Jønsson et al. 2015). In the con- tion (Esselstyn et al. 2009; Maestri et al. 2016; Pavan et al. 2016). text of the LOE rodents, the colonizing ancestor of Chrotomyini Thus, geographic scale and isolation likely influence the mech- may have been ecologically distinct enough to facilitate coexis- anism by which diversification occurs, as has been suggested in tence through divergent evolutionary trajectories. For example, the tropical Andes (Hutter et al. 2017). Despite the limited space comparative analysis of Floridian Quercus oaks suggests that, afforded to colonists of Luzon compared to continental radia- within ecological communities, individual species are less eco- tions, the patterns of LOE diversification were more consistent logically similar and more phylogenetically distant than would with a mechanism of allopatric speciation, where species diver- be expected by chance (Cavender-Bares et al. 2004). The phylo- sity is able to accumulate nearly constantly, offset slightly by genetic and ecological overdispersion reported by these authors background extinction rates. This conclusion is consistent with may extend more broadly to the island scale such that, for a the observation that Luzon, as a volcanic oceanic island, has had lineage to successfully invade, it must be phenotypically distinct a markedly dynamic history: tectonic uplift, sea level fluctua- enough from other lineages to minimize competition. Preliminary tions, and volcanism have all contributed to substantial changes evidence to support this prediction stems from field observations in available land area and suitable habitat (Hall 2013; Heaney et al. suggesting the LOE rodent clades differ in their modal dietary and

EVOLUTION JULY 2018 1431 D. M. ROWSEY ET AL.

habitat preferences: members of Phloeomyini are primarily arbo- AUTHOR CONTRIBUTIONS real herbivores whereas chrotomyine species are predominantly DMR and SAJ designed the research; DMR, LRH,andSAJ collected DMR DMR LRH SAJ terrestrial animalivores (Heaney et al. 2016a). the data; analyzed the data; and , ,and wrote the manuscript. One additional aspect that is also worth considering is the po- tential impact the two Old Endemic clades have had on the diver- ACKNOWLEDGMENTS sification of the Luzon New Endemics. We did not include these We are grateful to A. Ferguson, J. Phelps, the late W. Stanley (FMNH), as more recent colonists in our analyses, because they lack species well as D. Lunde (USNM) for their assistance in lending specimens for diversity due to their relatively recent origin on Luzon. Never- use in this study. We would like to thank C. Martin for laboratory assis- theless, these recently colonizing species do exhibit substantial tance and P.-H. Fabre who provided Hapalomys delacouri and Abditomys latidens sequences. We also thank F. K. Barker and two anonymous ecological differences from the LOE rodents. As an example, the reviewers who provided comments on drafts of the manuscript. The Flo- New Endemics are typically found either in low elevation or dis- rence Rothman Fellowship and Dayton Wilkie Fellowship awarded to turbed habitats. Their distribution may be the result of competitive DMR by the Bell Museum of Natural History provided funding neces- exclusion by the Luzon Old Endemics, which are concentrated in sary to perform DNA extraction and sequencing for this project. DMR high elevation, primary montane forest habitats (Heaney et al. was also supported by the Wallace and Mary Dayton Fellowship awarded by the Bell Museum of Natural History. Field research that produced 2016a). The ecological disparity may otherwise stem from an on- most of the Philippines specimens utilized herein has been supported by going taxon cycle in which the two colonization regimes are in the Barbara Brown Fund for Mammal Research of the Field Museum, different stages: the Old Endemics may have achieved relative and the Negaunee Foundation. Field research permits were provided by stasis due to their comparatively ancient colonization and subse- the Biodiversity Management Bureau of the Philippine Department of Environment and Natural Resources. quent ecological specialization; in contrast, the New Endemics, due to recent colonization, represent an incipient taxon expan- DATA ARCHIVING sion (Ricklefs and Bermingham 2002). Whether the differences Sequence data are deposited in GenBank using accession numbers in habitat use is the result of active exclusion by the LOE ro- MH330617-MH330660. Nucleotide alignments and tree files are de- dents or habitat preferences among the Luzon New Endemics posited in TreeBase using accession number 22736. remains to be discovered, but could shed additional light on how incumbency alters macroevolutionary processes and community CONFLICT OF INTEREST STATEMENT assembly. The authors declare no conflicts of interest.

LITERATURE CITED Achmadi, A. S., J. A. Esselstyn, K. C. Rowe, I. Maryanto, and M. T. Abdullah. Conclusions 2013. Phylogeny, diversity, and biogeography of Southeast Asian spiny rats (Maxomys). J. Mamm. 94:1412–1423. The extent to which ecological interactions among species and Adkins, R. M., E. L. Gelke, D. Rowe, and R. L. Honeycutt. 2001. Molecu- clades influence macroevolutionary processes is an important lar phylogeny and divergence time estimates for major rodent groups: question for understanding the origination and maintenance of evidence from multiple genes. Mol. Biol. Evol. 18:777–791. biodiversity. Our study examined whether incumbency of another Akaike, H. 1974. A new look at the statistical model identification. IEEE Trans. Auto. Cont. 19:716–723. clade diminishes the rate at which secondary colonists diversify Alfaro, M. E., F. Santini, C. Brock, H. Alamillo, A. Dornburg, D. L. Rabosky, while also providing the most complete record of Luzon Old G. Carnevale, and L. J. Harmon. 2009. Nine exceptional radiations plus Endemic rodent phylogenetic diversity. We did not recover a sig- high turnover explain species diversity in jawed vertebrates. Proc. Nat. nal of incumbency-mediated diversification, but there is still the Acad. Sci. USA 106:13410–13414. Alhajeri, B. H., J. J. Schenk, and S. J. Steppan. 2016. Ecomorphological di- potential for inter-clade interactions to have shaped the diversi- versification following continental colonization in muroid rodents (Ro- fication of both. Given the complexity of natural systems, many dentia: Muroidea). Biol. J. Linn. Soc. 117:463–481. processes have likely contributed to the evolution of species diver- Alroy, J. 1996. Constant extinction, constrained diversification, and unco- sity to varying degrees over the history of Luzon Island, including ordinated stasis in North American mammals. Palaeogeo. Palaeoclim. incumbency effects, within-clade interspecific competition, and Palaeoecol. 127:285–311. Baldwin, B. G., and M. J. Sanderson. 1998. Age and rate of diversification of allopatric speciation through island and/or montane habitat hop- the Hawaiian silversword alliance (Compositae). Proc. Nat. Acad. Sci. ping (Heaney et al. 2016b). Nevertheless, we argue that ecological USA 95:9402–9406. opportunity does not appear to be the primary mechanism deter- Bennet, G. M., and P. M. O’Grady. 2013. Historical biogeography and ecolog- mining the generation of species among these mammals and pro- ical opportunity in the adaptive radiation of native Hawaiian leafhoppers (Cicadellidae: Nesophrosyne). J. Biogeogr. 40:1512–1523. pose that incumbency effects may only be detectable in spatially Benson, D. A., M. Cavanaugh, K. Clark, I. Karsch-Mizrachi, D. J. Lip- limited scales that have remained geologically and climatically man, J. Ostell, and E. W. Savers. 2013. GenBank. Nucleic. Acids. Res. static over evolutionary time. 41(Database issue):D36–D42.

1432 EVOLUTION JULY 2018 DIVERSIFICATION OF LUZON OLD ENDEMIC MURINES

Betancur-R., R., G. Ort´ı, A. M. Stein, A. P. Marceniuk, and R. A. Pyron. 2012. Heaney, L. R., D. S. Balete, M. R. M. Duya, M. V. Duya, S. A. Jansa, S. J. Apparent signal of competition limiting diversification after ecologi- Steppan, and E. A. Rickart. 2016b. Doubling diversity: a cautionary tale cal transitions from marine to freshwater habitats. Ecol. Lett. 15:822– of previously unsuspected mammalian diversity on a tropical oceanic 830. island. Frontiers Biogeog. 8.2:e29667. Blackburn, D. C., C. D. Siler, A. C. Diesmos, J. A. McGuire, D. C. Cannatella, Hoehna, S. 2013. Fast simulation of reconstructed phylogenies under global, and R. M. Brown. 2013. An adaptive radiation of frogs in a southeast time-dependent birth-death processes. Bioinformatics 29:1367–1374. Asian island archipelago. Evolution 67:2631–2646. Hurvich, C. M., and C.-L. Tsai. 1989. Regression and time series model Borregaard, M. K., I. R. Amorim, P.A. V.Borges, J. S. Cabral, J-M. Fernandez- selection in small samples. Biometrika 76:297–307. Palacios, R. Field, L. R. Heaney, H. Kreft, T. J. Matthews, J. M. Olesen, Hutter, C. R., S. M. Lambert, and J. J. Wiens. 2017. Rapid diversification and et al. 2017. Oceanic island biogeography through the lens of the General time explain amphibian richness at different scales in the tropical Andes, Dynamic Model: assessment and prospect. Biol. Rev. 92:830–853. Earth’s most biodiverse hotspot. Am. Nat. 190:828–843. Bouckaert, R., J. Heled, D. Kuhnert,¨ T. Vaughan, C.-H. Wu, D. Xie, M. Jansa, S. A., F. K. Barker, and L. R. Heaney. 2006. The pattern and timing of A. Suchard, A. Rambaut, and A. J. Drummond. 2014. BEAST 2: a diversification of Philippine endemic rodents: evidence from mitochon- software platform for Bayesian evolutionary analysis. PLoS Comp. Biol. drial and nuclear gene sequences. Syst. Biol. 55:73–88. 10:e1003537. Jansa, S. A., F. K. Barker, and R. S. Voss. 2014. The early diversification his- Burns, K. J., A. J. Shultz, P. O. Title, N. A. Mason, F. K. Barker, J. Klicka, tory of Didelphid marsupials: a window into South America’s “splendid S. M. Lanyon, and I. J. Lovette. 2014. Phylogenetics and diversifica- isolation”. Evolution 68:684–695. tion of tanagers (Passeriformes: Thraupidae), the largest radiation of Jansa, S. A., S. M. Goodman, and P. K. Tucker. 1999. Molecular phylogeny Neotropical songbirds. Mol. Phy. Evol. 75:41–77. and biogeography of the native rodents of Madagascar (Muridae: Ne- Case, T. J. 1991. Community collapse in metapopulation models with inter- somyinae): a test of the single-origin hypothesis. Cladistics 15:253–270. species competition. Biol. J. Linn. Soc. 42:239–266. Jansa, S. A., and M. Weksler. 2004. Phylogeny of muroid rodents: relation- Cavender-Bares, J., D. D. Ackerly, D. A. Baum, and F. A. Bazzaz. 2004. ships within and among major lineages as determined by IRBP gene Phylogenetic overdispersion in Floridian oak communities. Am. Nat. sequences. Mol. Phylogenet. Evol. 31:256–276. 163:823–843. Jønsson, K. A., J.-P. Lessard, and R. E. Ricklefs. 2015. The evolution of Derryberry, E. P., S. Claramunt, G. Derryberry, R. T. Chesser, J. Cracraft, morphological diversity in continental assemblages of passerine birds. A. Aleixo, J. Perez-Em´ an,´ J. V. Remsen, Jr., and R. T. Brumfield. Evolution 69:879–889. 2011. Lineage diversification and morphological evolution in a large- Justiniano, R., J. J. Schenk, D. S. Balete, E. A. Rickart, J. A. Esselstyn, scale continental radiation: the neotropical ovenbirds and woodcreepers L. R. Heaney, and S. J. Steppan. 2015. Testing diversification models (Aves: Furnariidae). Evolution 65:2973–2986. of endemic Philippine forest mice (Apomys) with nuclear phylogenies Dos Reis, M., P. C. J. Donoghue, and Z. Yang. 2014. Neither phylogenomic across elevational gradients reveals repeated colonization of isolated nor palaeontolological data support a Palaeogene origin of placental mountain ranges. J. Biogeogr. 42:51–64. mammals. Biol. Lett. 10:1–4. Kimura, Y., M. T. R. Hawkins, M. M. McDonough, L. L. Jacobs, and L. Drake, J. A. 1991. Community-assembly mechanics and the structure of an J. Flynn. 2015. Corrected placement of Mus-Rattus fossil calibration experimental species ensemble. Am. Nat. 137:1–26. forces precision in the molecular tree of rodents. Nat. Sci. Rep. 5:1–9. Drummond, A. J., S. W. Y. Ho, M. J. Phillips, and A. Rambaut. 2006. Relaxed Kyriazis, C. C., J. M. Bates, and L. R. Heaney. 2017. Dynamics of ge- phylogenetics and dating with confidence. PLoS Biol. 4:e88. netic and morphological diversification in an incipient intra-island radi- Edgar, R. C. 2004. MUSCLE: multiple sequence alignment with high accuracy ation of Philippine rodents (Muridae: Bullimus). J. Biogeogr. 44:2585– and high throughput. Nucl. Ac. Res. 32:1792–1797. 2594. Esselstyn, J. A., R. M. Timm, and R. M. Brown. 2009. Do geological Lack, D. 1947. Darwin’s Finches. Cambridge University Press, Cambridge. or climatic processes drive speciation in dynamic archipelagos? The Lanfear, R., B. Calcott, S. Y. W. Ho, and S. Guindon. 2012. PartitionFinder: tempo and mode of diversification in southeast Asian shrews. Evolution combined selection of partitioning schemes and substitution models for 63:2595–2610. phylogenetic analyses. Mol. Biol. Evol. 29:1695–1701. Fukami, T., H. J. E. Beaumont, X. -X. Zhang, and P.B. Rainey. 2007. Immigra- Losos, J. B. 2011. Lizards in an Evolutionary Tree: Ecology and Adaptive tion history controls diversification in experimental adaptive radiation. Radiation of Anoles. Univ. of California Press, Berkeley, CA. Nature 446:436–439. Louette, G., and L. de Meester. 2007. Predation and priority effects in exper- Giarla, T. C., and S. A. Jansa. 2014. The role of physical geography and habitat imental zooplankton communities. Oikos 116:419–426. type in shaping the biogeographical history of a recent radiation of Maestri, R., L. R. Monteiro, R. Fornel, N. S. Upham, B. D. Patterson, and Neotropical marsupials (Thylamys: Didelphidae). J. Biogeogr. 41:1547– T. R. O. de Freitas. 2016. The ecology of a continental evolutionary 1558. radiation: is the radiation of sigmodontine rodents adaptive? Evolution Gillespe, R. 2004. Community assembly through adaptive radiation in Hawai- 71:610–632. ian spiders. Science 303:356–359. Mahler, D. L., L. J. Revell, R. E. Glor, and J. B. Losos. 2010. Ecological Goldberg, E. E., and R. Lande. 2007. Species’ borders and dispersal barriers. opportunity and the rate of morphological evolution in the diversification Am. Nat. 170:297–304. of Greater Antillean anoles. Evolution 64:2731–2745. Gu, X., Y. X. Fu, and W. H. Li. 1995. Maximum likelihood estimation of the Miller, M. A., W. Pfeiffer, and T. Schwartz. 2010. Creating the CIPRES heterogeneity of substitution rate among nucleotide sites. Mol. Biol., Science Gateway for inference of large phylogenetic trees. 2010 Gate- and Evol. 12:546–557. way Computing Environments Workshop, GCE 2010. New Orleans, Hall, R. 2013. The palaeogeography of Sundaland and Wallacea since the Late Louisiana. Jurassic. J. Limno. 72:1–17. Musser, G. G., and M. D. Carleton. 2005. Family Muridae. Pp. 1189–1531 in Heaney, L. R., D. S. Balete, and E. A. Rickart. 2016a. The Mammals of Luzon Mammal Species of the World: A Taxonomic and Geographic Reference. Island: Biogeography and Natural History of a Philippine Fauna. Johns 3rd ed. (Wilson, D. E., and D. M. Reeder, eds.) John Hopkins Univ. Press, Hopkins Univ. Press, Baltimore. Baltimore, Maryland.

EVOLUTION JULY 2018 1433 D. M. ROWSEY ET AL.

Nee, S., A. Ø. Mooers, and P. H. Harvey. 1992. Tempo and mode of evolu- Schenk, J. J., K. C. Rowe, and S. J. Steppan. 2013. Ecological opportunity and tion revealed from molecular phylogenies. Proc. Nat. Acad. Sci. USA incumbency in the diversification of repeated continental colonizations 89:8322–8326. by muroid rodents. Syst. Biol. 62:837–864. Nee, S., R. M. May, and P. H. Harvey. 1994. The reconstructed evolutionary Schluter, D. 2000a. The ecology of adaptive radiation. Oxford Univ. Press, process. Phil. Trans. Roy. Soc. B 344:305–311. Oxford. O’Leary, M. A., J. I. Bloch, J. J. Flynn, T. J. Gaudin, A. Giallombardo, N. Schluter, D. 2000b. Ecological character displacement in adaptive radiation. P. Giannini, S. L. Goldberg, B. P. Kraatz, Z.-X. Luo, J. Meng, et al. Am. Nat. 156:S4–S16. 2013. The placental mammal ancestor and the post-K-Pg radiation of Schluter, D., and P. R. Grant. 1983. Determinants of morphological patterns placentals. Science 339:662–667. in communities of Darwin’s finches. Am. Nat. 123:175–196. Pages,` M., P.-H. Fabre, Y. Chaval, A. Mortelliti, V. Nicolas, K. Wells, J. R. Schwarz, G. 1978. Estimating the dimension of a model. Ann. Stat. 6:461–464. Michaux, and V.Lazzari. 2016. Molecular phylogeny of Southeast Asian Skipwith, P. L., A. M. Bauer, T. R. Jackman, and R. A. Sadlier. 2016. Old but arboreal murine rodents. Zool. Script. 45:349–364. not ancient: coalescent species tree of New Caledonian geckos reveals Paradis, E., J. Claude, and K. Strimmer. 2004. APE: analyses of phylogenetics recent post-inundation diversification. J. Biogeogr. 43:1266–1276. and evolution in R language. Bioinformatics 20:289–290. Smith, M. F., and J. L. Patton. 1991. Variation in mitochondrial cytochrome b Parent, C. E., and B. J. Crespi. 2009. Ecological opportunity in adap- sequence in natural populations of South American Akodontine rodents tive radiation of Galapagos´ endemic land snails. Am. Nat. 174:898– (Muridae: Sigmodontinae). Mol. Biol. Evol. 8:85–103. 905. Stamatakis, A., P. Hoover, and J. Rougemont. 2008. A rapid bootstrap algo- Pavan, S. E., S. A. Jansa, and R. S. Voss. 2016. Spatiotemporal diversification rithm for the RAxML web servers. Syst. Biol. 57:758–771. of a low-vagility Neotropical vertebrate clade (short-tailed opossums, Stanhope, M. J., M. R. Smith, V. G. Waddell, C. A. Porter, M. S. Shivji, and Didelphidae: Monodelphis). J. Biogeogr. 43:1299–1309. M. Goodman. 1996. Mammalian evolution and the interphotoreceptor Pinto, G., D. L. Mahler, L. J. Harmon, and J. B. Losos. 2008. Testing the retinoid binding protein (IRBP) gene: convincing evidence for several island effect in adaptive radiation: rates and patterns of morphological superordinal clades. J. Mol. Evol. 43:83–92. diversification in Caribbean and mainland Anolis lizards. Proc. Roy. Soc. Stephens, P. R., and J. J. Wiens. 2003. Explaining species richness from B 275:2749–2757. continents to communities: the time-for-speciation effect in Emydid Pybus, O.G., and P.H. Harvey. 2000. Testing macro-evolutionary models us- turtles. Am. Nat. 161:112–128. ing incomplete molecular phylogenies. Proc. R. Soc. B 267: 2267– Steppan, S. J., R. M. Adkins, P. Q. Spinks, and C. Hale. 2005. Multigene 2272. phylogeny of the old world mice, Murinae, reveals distinct geographic R Core Team. 2015. R: A language and environment for statistical computing. lineages and the declining utility of mitochondrial genes compared to R Foundation for Statistical Computing, Vienna, Austria. nuclear genes. Mol. Phylo. Evol. 37:370–388. Rabosky, D. L. 2006. LASER: a maximum likelihood toolkit for detecting Steppan, S. J., and J. J. Schenk. 2017. Muroid rodent systematics: 900-species temporal shifts in diversification rates from molecular phylogenies. Evo- tree reveals increasing diversification rates. PLoS One 12:e0183070. lutionary Bioinformatics Online 2:273–276. Sutherland, J. P. 1974. Multiple stable points in natural communities. Am. Rabosky, D. L. 2012. Testing the time-for-speciation effect in the assembly Nat. 108:859–873. of regional biotas. Methods Ecol. Evol. 3:224–233. Tan, J., Z. Pu, W. A. Ryberg, and L. Jiang. 2012. Species phylogenetic relat- Rabosky, D. L. 2014. Automatic detection of key innovations, rate shifts, and edness, priority effects, and ecosystem functioning. Ecology 93:1164– diversity-dependence on phylogenetic trees. PLoS One 9:e89543. 1172. Rabosky, D. L., M. Grundler, C. Anderson, P. Title, J. J. Shi, J. W. Brown, Thomas, H., S. Sen, M. Khan, B. Battail, and G. Ligabue. 1982. The Lower H. Huang, and J. G. Larson. 2014. BAMMtools: an R package for the Miocene fauna of Al-Sarrar (Eastern Province, Saudi Arabia). Atlal analysis of evolutionary dynamics on phylogenetic trees. Methods Ecol. 5:109–136. Evol. 5:701–707. Tokita, M., W. Yano, H. F. James, and A. Abzhanov. 2016. Cranial shape and Rabosky, D. L., and I. J. Lovette. 2008. Density-dependent diversification in evolution in adaptive radiations of birds: comparative morphometrics of North American wood warblers. Proc. R. Soc. B 275:2363–2371. Darwin’s finches and Hawaiian honeycreepers. Phil. Trans. R. Soc. B Reddy, S., A. Driskell, D. L. Rabosky, S. J. Hackett, and T. S. Schulenberg. 372:1–17. 2012. Diversification and the adaptive radiation of the vangas of Mada- Wainwright, P. C., W. L. Smith, S. A. Price, K. L. Tang, J. S. Sparks, L. A. gascar. Proc. R. Soc. B 279:2062–2071. Ferry, K. L. Kuhn, R. I. Eytan, and T. J. Near. 2012. The evolution Revell, L. J. 2012. Phytools: an R package for phylogenetic comparative of pharyngognathy: a phylogenetic and functional appraisal of the pha- biology (and other things). Methods Ecol. Evol. 3:217–223. ryngeal jaw key innovation in Labroid fishes and beyond. Syst. Biol. Ribas, C. C., R. G. Moyle, C. Y. Miyaki, and J. Cracraft. 2007. The assembly 61:1001–1027. of montane biotas: linking Andean tectonics and climatic oscillations to Whittaker, R. J., and J. M. Fernadez-Palacios. 2007. Island Biogeography: independent regimes of diversification in Pionus parrots. Proc. R. Soc. Ecology, Evolution, and Conservation. Oxford Univ. Press, Oxford. B 274:2399-2408. Yoder, J. B., E. Clancey, S. Des Roches, J. M. Eastman, L. Gentry, W. Godsoe, Ricklefs, R. E., and E. Bermingham. 2002. The concept of the taxon cycle in T. J. Hagey, D. Jochimsen, B. P. Oswald, J. Robertson, et al. 2010. biogeography. Glob. Ecol. Biogeogr. 11:353–361. Ecological opportunity and the origin of adaptive radiations. J. Evol. Rowe, K. C., A. S. Achmadi, and J. A. Esselstyn. 2016. Repeated evolution Biol. 23:1581–1596. of carnivory among Indo-Australian rodents. Evolution 70:6. Yule,G. U. 1924. A mathematical theory of evolution based on the conclusions Rowe, K.C., M. L. Reno, D. Richmond, R. M. Adkins, and S. J. Steppan. of Dr. J. C. Willis. Phil. Trans. R. Soc. B 213:21–87. 2008. Pliocene colonization and adaptive radiations in Australia and New Guinea (Sahul): multilocus systematics of the Old Endemic rodents Associate Editor: I. Lovette (Muroidea: Murinae). Mol. Phylogenet. Evol. 47:84–101. Handling Editor: Mohamed A.F. Noor

1434 EVOLUTION JULY 2018 DIVERSIFICATION OF LUZON OLD ENDEMIC MURINES

Supporting Information Additional supporting information may be found online in the Supporting Information section at the end of the article.

Figure S1. Gene MCC trees generated for BRCA1, CYTB, FGB7, GHR, IRBP, and RAG1. Figure S2. BAMM maximum a posteriori (MAP) rate configuration based on MCC tree. Figure S3. BAMM 66.3% credible shift set of different rate configurations ordered by their sampling probabilities (f). Table S1. Full specimen matrix for phylogenetic analyses. Cells indicate GenBank numbers when available. Table S2. Clade-specific sampling probabilities used in our BAMM analysis accompanied by source(s) used for justifying the number of species in the clade.

EVOLUTION JULY 2018 1435